Re: [HACKERS] 6.6 release - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: [HACKERS] 6.6 release
Date
Msg-id 199912141627.LAA14646@candle.pha.pa.us
Whole thread Raw
In response to Re: [HACKERS] 6.6 release  (Vince Vielhaber <vev@michvhf.com>)
List pgsql-hackers
> On Fri, 10 Dec 1999, Thomas Lockhart wrote:
> 
> > imo the *only* reason we are tempted to do more major releases is that
> > we are too lazy/understaffed/sensible (you pick it) to support
> > multiple db formats for our compiled code. Other commercial DBs don't
> > release often, and they don't include big improvements, but they *do*
> > include support for multiple db formats/schemas in their product, so
> > you aren't forced into an initdb for each release. Instead they
> > include klugy workaround code to allow reading older formats with the
> > newer version.
> 
> Then why don't we come up with something to autoconvert the user's
> databases without having to dump/initdb/reload?  Or is that just
> not feasable (impossible's an answer I'd find hard to believe, but
> more trouble than it's worth is understandable).

System table changes often make that difficult.  pg_upgrade does most of
what we want by keeping the disk tables and allowing initdb.  If we
don't change the on-disk structure of user tables, pg_upgrade allows
quick upgrades.  Not sure 7.0 will allow the use of pg_upgrade.  6.5 did
not because the on-disk table structure changed with MVCC.

--  Bruce Momjian                        |  http://www.op.net/~candle maillist@candle.pha.pa.us            |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026
 


pgsql-hackers by date:

Previous
From: Bruce Momjian
Date:
Subject: Re: [HACKERS] Volunteer: Large Tuples / Tuple chaining
Next
From: Tom Lane
Date:
Subject: Re: [HACKERS] Re: [PATCHES] createdb/dropdb fixes